Professional Documents
Culture Documents
No Right Answer PDF
No Right Answer PDF
Copyright Information
NO RIGHT ANSWER?*
RONALD DWORKIN**
Responding to his earlier essays, where it was argued that hard cases hare right
answers, ProfessorDworkin's critics have maintainedthat .ases often arise in which
there is no right answer, and thatjudges as a consequence exercise discretion. Pro-
fessor Dworkin now details a refutation of two versions of the claim that there is no
right answer in a case at law. He first argues that an understandingof dispositirce
legal concepts, which link individual rights with official duty, defeats the claim that
there is a logical space between affirnatice and negative judicial results such that
neither may be right. He then assembles and dismantles argurwntsfrom vagueness.
positivism, and controversy, each of which purports to show that there is sufficient
indeterminacy in judicial decisions to prcclude demonstrable rightness of result. Fi-
nally, a rarity thesis is sketched to show that ecen if there are some cases in which
there is no right answer, they are sufficiently rare in a mature legal system as to be
exotic.
terms of the contract, whose validity Tim contests. Shall we say that
the judge must look for the right answer to the question of whether
Tom's contract is valid, even though the community is deeply divided
about what the right answer is? Or is it more realistic to say that
there simply is no right answer to the question?
That issue is central to a large number of controversies about
what law is. It has been debated under many titles, including the
question of whether judges always have discretion in hard cases, and
whether there are what some legal philosophers call "gaps" in the
law. I now wish to defend the unpopular view, that in the cir-
cumstances just described the question of Tom's contract may well
have a right answer, against certain arguments on which its oppo-
nents knowingly or unknowingly rely. I shall also try to show what
sense there is in the no-right-answer thesis, and- why the occasions
when a legal question has no right answer in our own legal system
may be much rarer than is generally supposed. I shall begin, how-
ever, by insisting upon a clarification of the issue that removes a
troublesome ambiguity.
Certain legal concepts, like the concept of a valid contract, of
civil liability, and of a crime, have the following characteristic: If the
concept holds in a particular situation, then judges have a duty, at
least prima facie, to decide some legal claim one way; but if the con-
cept does not hold, then judges have a duty, at least prima facie, to
decide the same claim in the opposite way. I shall call such concepts
"dispositive" concepts. Lawyers seem to assume, in the way they talk
and argue, what we might call the "bivalence thesis" about dispositive
concepts: that is, that in every case either the positive claim, that the
case falls under a dispositive concept, or the opposite claim, that it
does not, must be true even when it is controversial which is true.
Lawyers seem to assume, for example, that an exchange of promises
either does or does not constitute a valid contract. If it does, then
judges have at least a prima facie duty to enforce these promises if so
requested within their jurisdiction; but if it does not, then they have
at least a prima facie duty not to do so on contractual grounds.
Lawyers seem to assume that a particular person is either liable in
law for the damage his act has caused or he is not; if he is, then
judges have a duty to hold him in damages, but if he is not, then
they have a duty not to. They seem to assume that a particular piece
of conduct, taking into account intention and circumstances, either
constitutes a crime or it does not; if it does, and the actor has no
other defense, then the judge (or jury) has a duty to hold him guilty;
but if it does not, then the judge (or jury) has a duty to find him
innocent.
I See Dummett, Truth, in PHILOSOPHICAL LOGIC 64-66 (P. Strawson ed. 1967).
II
THE FIRST VERSION
We can easily imagine a legal system such that if anyone claimed
that there is always a right answer to the question of whether judges
have a duty to enforce an exchange of promises, or to refuse to en-
force the exchange, he would be making a mistake of the sort the first
version supposes. Even under our own law, after all, there are many
decisions that a judge has no duty to make either way. That is so, for
example, when the plaintiff requests an early adjournment on some
particular day and the defendant asks that the request be denied. It is
also so when the defendant has been convicted of a crime for which
the statute provides a sentence of from three to five years, and the
prosecution asks for the maximum, while the defense asks for the
minimum, sentence. The concept of duty provides a space between
the proposition that the judge has a duty to decide one way and the
proposition that he has a duty to decide another way; this space is
occupied by the proposition that he has no duty to decide one way or
the other, but rather a permission or, as lawyers say, a "discretion,"
to decide either way.
equivalences in meaning hold, then the first version of the thesis fol-
lows in a straightforward way. Since there is space between the two
propositions about judicial duty, and since the two propositions about
contracts mean the same thing as the propositions about judicial duty,
there must be space between the two latter propositions as well.
This argument would be impeccable if the semantic theory on
which it is based, that propositions of law are equivalent in meaning
to propositions about official duties, were sound. But that theory is
not sound. There must be some differences in meaning between the
proposition that a contract is valid and the proposition that judges
have a duty to enforce the promises that compose the contract, be-
cause the former statement is ordinarily taken as providing an argu-
ment for the latter, not simply as a question-begging restatement of
it. If there is a conceptual, and not simply a contingent, connection
between dispositive concepts and legal rights and duties, there is also
a conceptual, and not merely a contingent, connection between such
concepts and the types of events they report. If a lawyer says that his
client has a right to win a judgment because the contract on which he
sues is valid, or because the contract on which be is being sued is
invalid, he indicates his readiness to make certain sorts of arguments
rather than others, to point to facts having to do with offer, accep-
tance, capacity, illegality, or mistake rather than to other sorts of
facts, to support his client's claim. The semantic theory which simply
translates statements about contracts into statements about official
duties therefore obscures the interesting and distinctive role of dis-
positive concepts in legal argument. These concepts provide a special
kind of bridge between certain sorts of events and the conclusory
claims about rights and duties that hold if these events can be dem-
onstrated to have occurred. They both designate tests for conclusory
claims and insist that if the tests they designate are not met, then the
opposite conclusory claim, not simply the denial of the first, holds
instead. The need for concepts having that function in legal argument
arises because the concepts of right and duty in which conclusory
claims are framed are structured, that is, because there is space be-
tween the opposite conclusory claims. The function is the function of
denying that the space thus provided may be exploited by rejecting
both the opposing claims. Dispositive concepts are able to fill this
function just because the first version of the no-right-answer thesis is
false; if there were space between the propositions that a contract is
and is not valid, that concept could not close the space provided by
the concepts of right and duty.
The correct analogy, on this account of the matter, is not be-
tween dispositive legal concepts and factual events in a game, like a
contract is valid." But in that case no reason has been given to sup-
pose that "The law provides that Tom's contract is valid" and "The
law provides that Tom's contract is not valid" may both be false. It is
not evident that "Legally, Tom's contract is valid" and "Legally,
Tom's contract is not valid" may both be false. That is what the first
version must prove, not presuppose. If it strikes someone as evident
that "The law provides that the contract is valid" and "The law pro-
vides that the contract is not valid" may both be false, this is because
he personifies "the law"; that is, he takes it to be like a person who
may provide (p) or (-p) or neither. But the law is not a person.
Perhaps, however, the proposal is based, not on that redun-
dancy, but on a more ambitious semantics, which holds that ordinary
propositions of law have the same meaning as propositions about what
some person or institution has said. "The law provides that Tom's
contract is valid" may be read, on this understanding, as "Appropriate
authorities have decreed some rule according to which contracts like
Tom's must be enforced" or something of the sort. It may certainly
be false that appropriate authorities have decreed either that rule or a
rule requiring the opposite. But it is hardly evident that "Tom's con-
tract is not valid" means the same thing as "Authorities have enacted
some rule according to which the contract is not valid" (or that "Tom
is not guilty of a crime" means the same thing as "Authorities have
decreed some rule according to which what Tom did is not a crime").
On the contrary, that seems wrong. One strong argument against it is
just the fact that "Tom's contract is not valid" seems to be the nega-
tion of "Tom's contract is valid" (and "Tom is not guilty of a crime"
the negation of "Tom is guilty of a crime"). So the argument under
consideration (on this second interpretation as well as on the first) is
not an argument for the first version of the no-right-answer thesis; it
rather presupposes that thesis.
I shall mention one more apparent argument that the defender of
the first version might urge, which we might call the argument from
realism. He might say that my description of the function of disposi-
tive concepts must be wrong, because if it were right legal practice
would be grossly unrealistic in the following way: If we look to the
actual tests the law provides for claims about the validity of contracts,
we see that in fact there is sometimes no right answer to the question
of whether these tests are met in a particular case. Since there may
be no right answer to the question whether an agreement is sacrile-
gious or not, for example, there can be no right answer to the ques-
tion whether Tom's contract is valid or invalid, whether lawyers think
there is a right answer or not. This kind of indeterminacy occurs with
such frequency that it would be unrealistic and indeed perverse for
III
2 See Dworkin, Hard Cases, 88 HARV. L. REv. 1057 , 1085-87 (1975) [hereinafter Dwor.
kin, Hard Cases], reprinted in R. DWORKIN, TAKING RIGHTS SERIOUSLY 81, 107-10 (1977).
' V's argument assumed bivalence between "is true" and "is not true." Can R deny this,
and claim that "is true" is itself vague? V's argument that vagueness produces indeterminacy
relied on a distinction between "x is not '"and "it is not true that x is 4." That distinction is
necessary to his claim that "x is (b," and "x is not 4," might neither be true without either being
false. We can grasp the distinction only if we have independent criteria for asserting that some-
thing is "0k" and asserting that it is not. (I mean, by "independent," that the criteria for assert-
ing the one is not just the absence of the criteria for asserting the other.) Otherwise we could
make no sense of the idea that our criteria may not be satisfied for asserting either. The alleged
vagueness of "0k" consists in this independence of criteria. But can we distinguish in this way
between (1) "p is not true" and (2) "it is not true that p is true"? (1) says (on the analysis just
described) that the criteria for asserting (p) are not met. ftdoes not say that the criteria for
asserting (-p) are met. But (2) seems to say nothing more than the same thing, that Is, that tie
criteria for asserting (p) are not met. What- more or less might it be taken to claim? But if(1)
and (2) do not make different claims, then "is true" cannot be shown to be vague, at least on V's
theory of vagueness. The reader may feel that R has been cheated by this argument, After all,
the circumstance that R called attention to might well arise, for all this complex argument,
Someone told that if it is not true that a contract is sacrilegious he is to treat the contract as not
sacrilegious may still find himself in the difficulty of being uncertain whether it is not true that
the contract before him is sacrilegious. I agree. But that is a problem for V, not for my answer
to R. Someone defending the bivalence thesis I described earlier may say that every contract is
either sacrilegious or not, though it may be unclear which, and reasonable men may differ. V
must show that this claim is wrong, because the proposition that a contract is sacrilegious may
be neither true nor false. R's practical problem provides (I think) an embarrassment for V's
whole approach, at least if it is taken to be an argument for the second version of the no-right-
answer thesis.
4 In this essay I am concerned only with showing that legal positivism, even if it is true,
does not provide a good argument for the second version of the no-right-answer thesis. This
paragraph suggests an argument against positivism itself (it is, in fact, one way of stating what I
have in various lectures called the "simple-minded argument" against positivism). I do not pro-
pose to pursue that argument in this essay, but it might be useful to notice these points: (1) The
argument, as presented here, fails against what I called in the text semantic positivism. It fails
against, for example, a form of positivism that claims that "Tom's contract is valid" sicans that
judges have a duty to enforce the contract, and the proposition that "Tom's contract Is not
valid" means that judges have a duty not to enforce it. (But semantic positivism is indefensible.)
(2) The argument also fails against a form of positivism that sustains the following claims. Prop-
ositions of law can be divided into two classes, which may be called inherently positive (or
inherently mandatory or something of the kind) and inherently negative (or inherently pennis-
sive, etc.), such that, for every proposition of law and its negation, one is inherently positive
and the other inherently negative. If that is so. then a form of positivism can be defended -hich
argues that a positive proposition of lawv is truth-functionally equivalent to some statement about
lawmaking acts, so that, for example, it is true if and only if the sovereign has so commanded.
but that this is not so for negative propositions of law, which may be true just in virtue of the
failure of the sovereign to command the related positive proposition. But notice that this form of
positivism presupposes a kind of reductionism. It supposes. that is. that all propositions of law
that do not on the surface assert or deny duties or permissions can be translated, %ith no
change or loss in meaning, into propositions that do. It also supposes that when this reduction is
carried through, each proposition so reduced wvill belong to an opposite class from that to vhich
its negation is reduced, rather than, for example, each being seen to be (at bottom) claims of
permission that cannot, as a matter of law, both be true. It supposes. further. that the norma-
tive proposition it expresses, which is that whatever is not prohibited is permitted, is a fair
description of legal practice. This assumption may be reasonable in cases in %hich the law
intervenes on a clean slate, as when legal rules for property are provided for a community
which has no (pre-legal) scheme of ownership. It is unreasonable when same area of law de-
velops step-by-step rather than by deploying and then refining some all-eimbracing principle. as
in the case, for example, of the development of large parts of the law of negligence. I do not
regard these brief remarks as effective arguments against a canonical positivefnegaItive (or man-
datorylpermissive) distinction, but only as a reminder of the difficulties such a distinction must
surmount.
rule, or something of that sort. This form of positivism does not pre-
suppose the first version of the no-right-answer thesis, because it
does not suggest that there is any conceptual space, within the in-
stitution of law, between any proposition and its apparent negation. It
does not suppose that the proposition that a contract is valid and the
proposition that it is not valid may both be false. But it does support
the second version of the thesis, because it shows how a particular
proposition may be neither true nor false, not because of some
vagueness or open texture in canonical language, but because the
ground rules of the legal enterprise, like the ground rules of the liter-
ary enterprise I described have that consequence.
We must now notice that this form of positivism differs from
more familiar forms in one important respect. Orthodox positivism, in
each of its forms, claims some sort of conceptual connection between
lav and the particular act or acts designated by the theory as the
distinctive law-creating act. For an Austinian positivist, for example,
the fact that law is the command of the sovereign is not simply the
consequence of the particular legal practices in some countries. It is,
on the contrary, constitutive of the very idea of law. But the new
version of positivism I constructed, based on the analogy of the liter-
ary game, does not permit the positivist so global a claim. He must
be content to say that (as it happens) the citizens and officials in a
particular jurisdiction follow ground rules about the assertion and de-
nial of legal propositions such that no such proposition may be as-
serted unless a sovereign has made the appropriate command, or de-
nied unless he has made the contrary command, and that, for that
reason, there are propositions of lav that can neither be asserted nor
denied. But then his claim is not that there must be, in any legal
system, questions of law that have no right answer for that reason;
but only that there are such questions, for that reason, in the legal
system under consideration. He must concede the possibility, at
least, of other legal systems which follow very different ground rules
about the assertion and denial of propositions of law, and he must
also concede that questions of lav that do not have right answers in
the system he describes have right answers in those other systems,
even though no further commands or other lawmaking acts have
taken place there. 5 It is not difficult to imagine such other systems.
5 I hope that the -new" positivist will not make a different sort of claim. He might say that
a legal system exists only if citizens and officials follow the ground rules he has stipulated, and
that if they do not (but rather follow some different ground rules of the sort I describe In
succeeding paragraphs) then the arrangement does not count as a legal system. Ile could not
claim any justification in ordinary language for that piece of linguistic tyranny, so that his theory
would become simply an unprofitable stipulation, as if some student of literature claimed that
the different forms of literary criticism described in the next paragraph of the text were not
forms of literary criticism. He would fall into the same banality if he were to say that, although
a political arrangement might count as a legal system even if different ground rules were fol.
lowed, an answer to a question about what a court should do would count as a legal answer only
if it would be generated by his ground rules, whether or not it is generated by theirs.
provide a better (or worse) fit with the political theory that provides
the best justification for propositions of law ,already established.
The issue of whether there is a right answer to any particular
question of law will crucially depend upon which form of the legal
enterprise is in play. If it is like the first form of the literary exercise,
then the question of whether Tom's contract is valid will not have a
right answer on the simple facts I stipulated at the start of the essay.
But if it is like the third form, on the other hand, that question will
almost certainly have a right answer, because, for reasons I consider
more fully in the next section, it is very unlikelv that one answer will
not provide a better fit in the sense just described. If a positivist
wishes to argue that in cases like Tom's case there is no right answer,
so that judicial discretion must be exercised willy-nilly, then he must
show that our own legal practice is like the first form of the literary
exercise and not like the third. (I leave aside the question of whether
the latter would count as a positivistic account of law at all.) But
whether our system is more like the first than the third form is a
question of fact. So even if we accept the general account of law I
described, which holds that legal propositions are not directly true or
false of some external world, but are rather propositions whose asser-
tion or denial is licensed by ground rules that vary %%ith practice,
nothing follows, from that general theory of law, about the extent, if
any, to which the second version of the no-right-answer thesis is true
of any particular legal jurisdiction.
that it does not hold can be true. If reasonable lawyers can disagree
whether Sunday contracts are sacrilegious within the meaning of the
statute, because they hold different views about how statutes con-
taining vague terms should be interpreted or construed, then the
proposition that Tom's contract is valid cannot be demonstrated to be
true, even when all facts about what the legislators had in mind are
known or stipulated. Therefore, on the demonstrability thesis, it can-
not be true. But the same holds for the proposition that Torn's con-
tract is not valid. Since neither of these propositions can be true, and
since they are assumed to exhaust the range of possible answers, then
there is no right answer to the question.
The demonstrability thesis therefore provides a conclusive argu-
ment for the second version of the no-right-answer thtsis. But why
should we accept the demonstrability thesis? Anyone will accept it, of
course, who holds a strict form of empiricism in metaphysics. If we
believe that no proposition can be true except in virtue of some fact
that makes it true, and that there are no facts in the world but hard
facts, then the demonstrability thesis follows from that metaphysics.
The proposition could rationally be believed to be true, even though
its truth is not demonstrated when all the hard facts are known or
stipulated, only if there were something else in the world in virtue of
which it could possibly be true. But if there is nothing else, then the
proposition cannot rationally be believed to be true; the failure of
hard facts to make it true would have exhausted all hope of making it
true.
But if, on the other hand, we suppose that there is something
else in the' world beside hard facts, in virtue of which propositions of
law might be true, then the demonstrability thesis, in the fbrm I set
it out, must be false. Suppose, for example, there are moral facts,
which are not simply physical facts or facts about the thoughts or
attitudes of people. I do not mean that there are what are sometimes
called "transcendent" or "Platonic" moral facts; indeed I do not know
what these would be. I mean only to suppose that a particular social
institution like slavery might be unjust, not because people think it
unjust, or have conventions according to which it is unjust, or any-
thing of the sort, but just because slavery is unjust. If there are such
moral facts, then a proposition of law might rationally be supposed to
be true even if lawyers continue to disagree about the proposition
after all hard facts are known or stipulated. It might be true in virtue
of a moral fact which is not known or stipulated.
The demonstrability thesis, therefore, seems to depend upon ai
answer to the question of what there is. I shall not, in this essay, try
to make plausible the idea that moral facts exist, but I shall try to
support the idea that some facts beside hard facts do. I wish, for this
purpose, to consider again the third form of the literary exercise I
described in the last section. Participants assert a proposition about
David as true (or deny it as false) if that proposition provides a better
(or worse) fit than its negation with propositions already established,
because it explains in a more satisfactory way why David did what he
did, or said what he said, or thought what he thought, according to
the established propositions.
I do not mean to raise the question, through this story, of
whether fictitious persons are in some sense real so that all these
propositions may be said to be true of someone or something. I do
not mean to suggest, that is, that in addition to hard facts there are
facts like the fact that David Copperfield first read Hamnet at Salem
House. The literary exercise I imagine does not require that assump-
tion to make it a sensible exercise. But it does require the assump-
tion, I think, that there are facts of narrative consistency, like the fact
that the hypothesis that David had a sexual relationship with Steer-
forth provides a more satisfactory explanation of what he subsequently
did and thought than the hypothesis that he did not.
That is not, I take it, a hard fact. It is not the sort of fact that is
even in principle demonstrable by ordinary scientific methods. Since
no one ever did have just the history and character Dickens said
David did, we cannot provide ordinary arguments of probability,
even when all the histories of real people are known, that would
necessarily convince any rational man either to accept or reject the
hypothesis. In some cases, the argument will be so strong for a par-
ticular proposition, no doubt, that we should say that any participant
who did not agree with that proposition was simply incompetent at
the exercise. In other cases, we should not say this at all; we should
say that there is so much to be said on both sides that competent
participants might reasonably disagree.
Suppose that the exercise proceeds with fair success. The par-
ticipants often agree, and even when they disagree they understand
the arguments on both sides well enough to rank each set, for exam-
ple, in rough order of plausibility. Suppose now that an empiricist
philosopher visits the proceedings of the group, and tells them that
there are no such things as facts of narrative consistency or that, in
any case, there are no such facts when reasonable men can disagree
about what they are. He adds that therefore no one can have an),
reason to think, in response to the terms of the exercise, that the
argument that David had an affair with Steerforth is stronger than the
argument that he did not. Why should they be persuaded by what he
says? This case is not like Dummett's example of Charles's bravery I
vhy people hold beliefs about hard facts. It is unclear how he might
conceivably show this. Perhaps he might invent a machine which
would be able to predict, with great accuracy, what a participant's
belief would be with respect to any question about David that might
be asked, once highly specific information about the participant's
blood chemistry was programmed into the machine. It is, of course,
very speculative that if such a machine were built it 'vould yield such
predictions in the case of this literary exercise, but not also in the
case of, for example, astronomers who debate about the number of
Jupiter's moons. If the demonstrability thesis depends on the specula-
tion that the machine would yield positive results in the one case, but
not in the other, then it rests on very, shaky ground.
Let us assume, nevertheless, that such a machine could be built,
and that it would yield that discriminatory information about the
literary exercise. What follows? The philosopher might be justified in
concluding that the literary exercise was special in the following
sense: In many exercises, including the experimental sciences, par-
ticipants are trained to respond to their observations of the external
world in a way which, we suppose, increases our collective knowl-
edge of the world. In the literary exercise, participants are trained to
respond to certain questions of a highly specific form which, as the
machine is supposed to have proved, cannot be said to be questions
about the external world. The), are trained to subject their responses
to the disciplines of reflection and consistency, and then to make cer-
tain assertions that their training authorizes them to make on the au-
thority of these responses so disciplined. The exercise, conducted by
participants so trained, serves some purpose- perhaps recreational or
cultural-other than to increase our collective knowledge of the ex-
ternal world.
Suppose this distinction, or some more sophisticated version, can
in fact be made out between enterprises like astronomy and enter-
prises like literary games. That would be an important discovery, and
we should certainly wish to mark the distinction in some way. Sup-
pose a philosopher argues that, in consequence of the distinction, we
should not say that propositions asserted by participants in the liter-
ary exercise can be either true or false. If he explained that he
wished to mark the important distinction in this way, we might or
might not agree that the constraint he suggests is an appropriate way
to do this. But we should be careful to stipulate what must not follow
from the decision to restrict the use of "true" and "false" in that way.
It must not follow, for example, that the participants have no
reason to think one judgment of narrative consistency superior to
another when they disagree about which is superior. They still have
tract is not valid, but is false if that justification provides a better case
for that contrary proposition. There are important differences be-
tween the idea of consistency used in this account of legal reasoning
and the idea of narrative consistency used in the literary exercise.
Legal reasoning makes use of the idea of normative consistency which
is plainly more complex than, and may be thought to introduce
grounds for claims of subjectivism not present in, narrative consis-
tency. Nevertheless the comparison may help to explain why it is
sensible to suppose that there might be a right answer to the ques-
tion of whether Tom's contract is valid even when that answer cannot
be demonstrated.
The comparison is useful in another way as well. It helps us to
understand why, even though we reject the demonstrability thesis
and therefore reject the idea that there is no right answer whenever
the right answer is not demonstrable, it might nevertheless be sensi-
ble to say that there is no right answer to a question of law in certain
very special cases. In certain circumstances, even in the third form of
the literary exercise, it might be right for the participants to refuse to
assert either that David had some property or that he did not. Sup-
pose the question is raised whether David had type-A blood or not,
and there is no reason to think that a boy with that blood type would
be more likely to have had the history and character Dickens stipu-
lates than a boy with any other blood type. The proposition that
David had type-A blood is not vague; we can say that any historical
boy would either have had type-A blood or not, and that there is a
right answer to the question whether he did, even though we shall
never know. But the assertion conditions of the literary exercise for-
bid saying that of David; it seems more sensible, given these condi-
tions, to say that though the proposition that he had that blood type
is not true, the proposition that he did not is not true either. In such
a case the grounds for saying that there is no right answer to the
question are not based on any external criticism of the enterprise, or
on any external philosophical position like the demonstrability thesis.
The grounds are simply that that is the right response within the
terms of the enterprise itself. We may imagine a genuine controversy
within the enterprise as to whether, in any particular case, that is the
right response. One party may say that there is a reason for thinking
that boys like David would for that reason have been more likely to
have type-A blood and another that there is a reason for thinking that
they would more likely not, and a third thinking either that there
were no reasons either way, or that whatever reasons there were
were so equally balanced that no sensible discrimination could be
made.